savu.data.data_structures.preview module

class Preview(data_obj)

The Data class dynamically inherits from transport specific data class and holds the data array, along with associated information.

convert_indices(idx, dim)

convert keywords to integers.

get_data_obj()

get_integer_entries(plist)

Convert Savu preview syntax to python slicing (similar) syntax, by replacing Savu Built-in constants.

Parameters

plist (list) – A Savu data preview list to reduce the data dimensions.

Returns

A Savu preview list containing integers and no strings.

Return type

list

get_starts_stops_steps(key=None)

Returns preview parameter starts, stops, steps, chunks lists.

Keyword Arguments

key (str) – the list to return.

Returns

if key is none return separate preview_list components, where each list has length equal to number of dataset dimensions, else only the key list.

Return type

list(list(int))

set_preview(preview_list, **kwargs)

Reduces the data to be processed to a subset of the original.

Parameters

preview (list) – previewing parameters, where len(preview_list) equals the number of data dimensions.

Keyword Arguments

revert (bool) – revert input dataset to the original size after plugin processing.

Each preview_list element should be of the form start:stop:step:chunk, where stop, step and chunk are optional (defaults: stop = start + 1, step = 1, chunk = 1) but must be given in that order.

Note

start:stop[:step]

represents the set of indices specified by:

>>> indices = range(start, stop[, step])

For more information see range()

start:stop:step:chunk (chunk > 1)

represents the set of indices specified by:

>>> a = np.tile(np.arange(start, stop, step), (chunk, 1))
>>> b = np.transpose(np.tile(np.arange(chunk)-chunk // 2, (a.shape[1], 1)))
>>> indices = np.ravel(np.transpose(a + b))

Chunk indicates how many values to take around each value in range(start, stop, step). It is only available for slicing dimensions.

Warning

If any indices are out of range (or negative) then the list is invalid. When chunk > 1, new start and end values will be:

>>> new_start = start - int(chunk / 2)
>>> new_end = range(start, stop, step)[-1] + (step - int(chunk / 2))

accepted values:

Each entry is executed using eval() so simple formulas are allowed and may contain the following keywords:

• : is a simplification for 0:end:1:1 (all values)

• mid is int(shape[dim] / 2)-1

• end is shape[dim]